The expanded tool allows users to check images, videos and audio for digital watermarks, offering newsrooms another way to verify online content.
Google has expanded public access to its SynthID Detector, giving users worldwide a new way to check whether images, videos and audio contain digital watermarks associated with supported artificial intelligence systems.
The move comes as AI-generated and AI-edited media become increasingly common online, making it more difficult for journalists, publishers and the public to determine the origins of content circulating on social media and other digital platforms.
Google DeepMind announced the expansion on October 7, saying the tool was now available globally in English. The company said the detector supports media generated using Google AI and participating partners, including OpenAI, NVIDIA and Kakao, with Apple expected to join later.
How SynthID Detector Works
SynthID is Google’s digital watermarking technology. It embeds signals into supported AI-generated media that are designed to remain undetectable to the human eye or ear while still being identifiable by compatible detection technology.
Users can submit supported images, videos or audio files to the SynthID Detector to check for the watermark.
Google DeepMind Vice-President of Science and Strategic Initiatives Pushmeet Kohli said the expansion was intended to give people more context about the content they encounter online and help them make informed decisions about what to trust.
Google said its technology had been used to watermark more than 180 billion images and videos, as well as approximately 240,000 years of audio content. These figures were provided by the company.
Why the Tool Matters for Journalists
For news organisations, the expanded detector offers another option for examining suspicious or potentially AI-generated media before publication.
A newsroom receiving a dramatic photograph, an apparently authentic video or a purported audio recording could use the tool as one step in assessing the content’s origin.
However, detecting an AI watermark is not the same as proving that a story is true or false. A genuine event can be represented in AI-edited media, while fabricated content may contain no detectable watermark.
The detector also cannot establish where or when an image was captured, whether the people shown are correctly identified, or whether the accompanying claims are accurate.
Important Limitations
Google’s detector is not a universal AI detector. It is designed to identify SynthID watermarks in supported media, so content generated by systems that do not use compatible watermarking may not be identified.
A result showing no detectable watermark therefore does not prove that a file was created by a human or that it has not been manipulated. Similarly, detecting a watermark does not necessarily mean the entire file was generated from scratch by AI; the media may have been edited using an AI tool.
These limitations mean that journalists should combine watermark checks with other verification methods, including reverse-image searches, original-source confirmation, metadata examination where appropriate, location and date checks, and independent corroboration.
What This Means for OnPointGist
For OnPointGist, tools such as SynthID Detector can support the editorial verification process, particularly when assessing viral images, videos and audio submitted by readers or circulating on social media.
The appropriate approach is to treat detector results as evidence about a file’s possible origin—not as a final verdict on its authenticity or the truth of a news claim.
As AI technology advances, responsible reporting will increasingly depend on combining technical tools with conventional journalism: checking sources, confirming context, seeking independent evidence and clearly communicating uncertainty to readers.
OnPointGist will continue to follow developments in AI verification technology and their implications for journalism, digital media and public trust.
